1. Field of the Invention
The invention relates to a system and method for testing semiconductor devices and, more particularly, to such a system and method in which when failures are detected, decision circuitry determines whether it is more efficient to retest or repair the semiconductor device.
2. State of the Art
Typically, finished integrated semiconductor device assemblies include a die or dice that is attached to a lead frame and encapsulated with an encapsulant. Numerous expensive and time-consuming steps are involved in producing such semiconductor device assemblies. These steps may include the following: (1) forming dice on a wafer substrate, (2) testing the dice, (3) cutting dice from the wafer, (4) connecting a die or dice to a lead frame, (5) encapsulating the die or dice, lead frame, connecting wires, and any auxiliary circuitry, (6) performing bum-in and/or providing other stresses to the dice, and (7) testing the semiconductor device assembly at various stages of processing.
In semiconductor manufacturing, typically, the term "front-end" refers to the fabrication of semiconductor devices to the level of completed and tested wafers. The term "backend" refers to production stages of semiconductor devices occurring after the front-end and including such semiconductor device production stages as packaging, bum-in, testing, sorting, marking, and environmental testing.
When tested, a semiconductor device may have some failure due to various causes including, but not limited to, an internal defect in the die or chip, a bad bonding connection, or a bad connection between a lead finger and a probe or other test device. Failures in a completed semiconductor device assembly can prevent it from operating as intended. In spite of painstaking attention to detail, failures may be introduced at various levels of production. For example, defects in forming the die may cause a failure. It has been found, however, that some defects are manifest immediately, while other defects are manifest only after the die has been operated for some period of time.
"Burn-in" refers to the process of accelerating failures that occur during the infant mortality phase of component life in order to remove the inherently weaker semiconductor devices. The process has been regarded as critical for product reliability since the semiconductor industry began. There have been two basic types of burn-in. During the process known as "static" burn-in, temperatures are increased (or sometimes decreased) while only some of the pins on a test semiconductor device are biased. No data is written to the semiconductor device, nor is the semiconductor device exercised under stress during static burn-in. During "unmonitored dynamic" burn-in of a semiconductor device, temperatures are increased while the pins on the semiconductor device being tested are biased.
In recent years, as semiconductor device systems have grown in complexity, the need for more and more reliable components has escalated. This need has been met in two ways. First, in semiconductor device manufacturing processes where the manufacturing process technology has reached a level of maturity and stability, inherent manufacturing defects in the semiconductor device caused by contamination and process variation have been reduced. As a result, latent failures in the semiconductor device have been significantly reduced, resulting in lower field or usage failure rates. Further, more sophisticated methods of screening infant mortality failures in semiconductor devices have been developed to help minimize such failures.
To address these issues, an "intelligent" burn-in approach of the semiconductor device can be utilized. The term "intelligent burn-in", as used in this discussion, refers to the ability to combine functional, programmable testing with the traditional burn-in cycling of the semiconductor device under test while the semiconductor device is located in the same chamber.
Some semiconductor devices have internal test modes not accessible during normal operation. These test modes may be invoked on automatic test equipment (ATE) by applying a high voltage to a single pin. The semiconductor device is then addressed in a manner so as to specify the operating mode of interest. Operating modes, such as data compression, grounded substrate, and cell plate biasing can be enabled, thus allowing evaluation of operating characteristics of the semiconductor device and help in isolating possible failure mechanisms.
The electrical characterization data gathered from such tests is then used to identify the part of the circuit of the semiconductor device that appears to be malfunctioning, the possible location(s) of such malfunctions on the semiconductor device, and the most probable type or nature of the defect of the semiconductor device. To facilitate discussion and reporting, semiconductor device failures are often classified according to their electrical characteristics, commonly referred to as the failure mode. Typical classification of these modes include the following: single cell defect, adjacent cell defect, row failure, column failure, address failure, open pin, supply leakage, pin leakage, standby current leakage, and entire array failure (all dead cells).
In anticipation that some semiconductor devices will have defects, many semiconductor devices are designed with redundancies. In such semiconductor devices, a defective section of the semiconductor devices may be shut off and a redundant but properly operating section activated and used in place of the defective section. For example, typical integrated memory circuits include arrays of memory cells arranged in rows and columns. In many such integrated memory arrays, several redundant rows and columns are provided to be used as substitutes for defective rows or columns of memory. When a defective row or column is identified in the array, rather than treating the entire array as defective, a redundant row or column is substituted for the defective row or column. This substitution is performed by assigning the address of the defective row or column in the array to the redundant row or column such that, when an address signal corresponding to the defective row or column is received, the redundant row or column is addressed instead.
To make the substitution of the redundant row or column in the array substantially transparent to an operating system employing the memory circuit, the memory circuit may include an address detection circuit. The address detection circuit monitors the row and column addresses and, when the address of a defective row or column is received, enables or substitutes the redundant row or column in the array for the defective row or column.
One type of address detection circuit for memory-type semiconductor devices is a fuse-bank address detection circuit. Fuse-bank address detection circuits employ a bank of sense lines where each sense line corresponds to a bit of an address in the array of memory circuits. The sense lines are programmed by blowing fuses in the sense lines in a pattern corresponding to the address of the defective row or column in the array of memory circuits. Addresses are then detected by first applying a test voltage across the bank of sense lines. Then, bits of the address are applied to the sense lines. If the pattern of blown fuses precisely corresponds to the pattern of address bits, the sense lines all block current and the voltage accross the bank remains high. Otherwise, at least one sense line is conductive and the voltage falls. Thus, a high voltage indicates the programmed address has been detected while a low voltage indicates a different address has been applied.
Antifuses have been used in place of conventional fuses. Antifuses are capacitive-type structures that, in their unblown states, form open circuits. Antifuses may be "blown" by applying a high voltage across the antifuse. The high voltage causes the capacitive-type structure to break down, thereby forming a conductive path through the antifuse.
Failures detected during a test of a semiconductor device may be caused by an actual failure in the semiconductor device or through a failure in the test system or through an environmental cause. If there is an actual failure of the semiconductor device, it is typically most efficient to repair the tested semiconductor device, if possible, rather than retest the semiconductor device (which may be very time consuming). On the other hand, if the failure is due to the test system or an environmental cause, the identification of failures will not be accurate and repair will not be an efficient use of resources.
Accordingly, there is a need for a system and method that determines whether it is more efficient to repair or retest a semiconductor device when failures are detected in testing.